At present, there are two main phenomena in the classical measurement econometric research method: on the one hand, the same frequency data are used in the research process; on the other hand, many studies use low-frequency stock market data as the data index of the research object, which makes Asian stock market data has the same frequency as macro exogenous explanatory variables. The purpose of this paper is to solve the problem that the traditional co-frequency model cannot obtain the causal relationship between the macroeconomic explanatory variables and the fluctuations of emerging Asian stock markets due to the limitation of data frequency. This paper based on the fuzzy rationality hypothesis, using fuzzy mathematics and fuzzy statistical tools to improves the traditional GARCH-MIDAS model. Through the improved GARCH-MIDAS method, studied the fluctuation of emerging Asian stock markets. The results show that through the ability prediction for improved models, analyzed and found that both single-factor and multi-factor models have strong predictive power. By comparison, we can find that the multi-factor mixing model can better describe long-term component of price volatility in emerging Asian stock markets than the single factor mixing model.